Although up to 30% of individuals with ASD are minimally verbal, extremely few interventions can reliably produce significant improvements in speech output. Recently, our laboratory has developed Auditory-Motor Mapping Training (AMMT), a novel intonation-based intervention, which aims to facilitate speech output in minimally verbal children with ASD. This intervention involves the mapping of sounds to oral articulatory actions through intonation and bimanual motor activities. AMMT is built upon the musical strengths and preferences that have been observed in these children. Furthermore, associating sounds with actions engages an auditory-motor network of brain regions important for speech that has been reported to be dysfunctional in ASD. The overall aim of the study is to compare the efficacy of AMMT to that of a control therapy (CT) in facilitating speech output in a randomized controlled trial (RCT). To examine whether the core components of AMMT are responsible for the treatment effects, CT will omit the intonation and motor components. The accuracy of each child's speech output will be assessed multiple times before, during, and after treatment by independent coders. Our overall aim will also be complemented by two additional aims, in which we will examine whether frequency of AMMT affects treatment outcome, and whether pre-existing variability in language-related pathways and functional connections correlate with outcome, and is changed after treatment. Based on our preliminary data, we hypothesize that: (1) compared to CT, AMMT will result in significantly greater improvements in the accuracy of consonant and vowel productions, with generalization to items that are not trained during the therapy,and more appropriate words after therapy, (2) greater improvements will be observed in individuals who undergo higher frequency (compared to lower frequency) of AMMT, (3) baseline behavioral measures (e.g., cognitive, speech praxis, joint attention abilities) will be related to the degree of speech improvements in both AMMT and CT, and (4) the degree of variability infunctional and structural brain connections will be related to AMMT treatment outcome, and (5) intensive AMMT treatment will lead to increased functional connections in the auditory-motor network. By using a combination of behavioral and brain imaging measures, the proposed study will examine the efficacy of a novel intervention, and enhance our understanding of the underlying neural mechanisms that underlie the therapy effects.